Did nothing, even less of a change than that of the VC viewing. The fences were easier to watch for changes while rocking the plane on the handbrake with the throttles showed no difference in the microscopic fence posts half a mile away.

...What I did find with the BN2 Islander is that: DX9 "Antialiasing - Setting = 32xS [Combined: 2x2 SS + 8x MS]" is marginally better than DX10's "Antialiasing - Transparency Supersampling = 8x Sparse Grid Supersampling". Applying 1.5xDSR improved the DX10 view of the BN2, albeit with the smaller menu and text.

That's fine that the thread took a different turn. While its off topic and there's dx10 discussion, can anyone point me in the right direction on how to convert dx9 to dx10 as far as settings for inspector, fsx, Cfg file. Any links appreciated.

This is best guide on moving to DX10 there is:

http://www.sim-outhouse.com/sohforums/showthread.php?88271-DX10-quot-How-to-quot-Guide-a-new-home!

Been using the Steve's DX10 fixer since it first came out, and months before that his modded DX10 shaders, which are available from the AVSIM library.  Try the free shaders first to see if it's what you want to do.  The fixer app adds these features on top of the free shaders: repairing "DX10 crippled" addons, enabling DX10 effects for legacy aircraft, tuning DX10 bloom, shadow and water effects, cloud graphics clipping and culling and other stuff which I've yet to investigate.  Enjoy.

how do you apply DSR with new driver update?

From the infamous link...

DSR: Use the NVidia Control Panel, Manage 3D Settings, Global Settings, DSR - Factors, and DSR - Smoothness. Check the desired multiplier boxes, and set the desired smoothness (quality of downscaling). Then in FSX full screen game mode, choose the increased custom resolution in the Display Settings. Use the window mode switching ALT+ENTER key combination twice to invoke the new setting. The screen text becomes smaller and the larger starting resolution is scaled onto the monitor resolution to produce the anti-aliased effect. DSR is not an alternative to other AA techniques, rather it enhances their results. Filtering will still be required to improve textures into the far distance. Aircraft models requiring Sparse Grid Supersampling in DX10 and DX11, still require a measure of that technique for acceptably smooth results overall.

 

Often after updating the drivers we need to refresh the Profile - also from the link...

Make a new profile for NI: What I do is use the NVidia GPU Control Panel, Manage 3D Settings, select FSX (or Prepar3D), press the "Restore" button and "Apply". Next I set "Enhance the application setting" and press "Apply", this is simply to help create a fresh NI profile. If we now go into NI there is a new FSX profile we can select and set our desired values. We can start this process by applying Restore to the Global Settings, if we want to create all new profiles. Usually it's best to start fresh with each version of driver or NI.

 

 

We only need to change the desktop resolution if we want to use FSX in windowed mode. Remember we need to use ALT+ENTER twice to swap out the fullscreen setup so that the NVidia driver can calculate the sizes.

8xS? I've been going mad with settings. After viewing the dropdowns over one thousand times, only then I noticed that I was mistaking the 8xS setting Bert mentioned and thinking 8xSQ was the next "combined method" in the dropdown, but you meant "1x2 SS + 2x MS", guys I apologise for that frustrating time. But in the end, still not happy. After a lot of testing with various VCs zoomed right out and eye-point moved right back, I can display a minute bit of pixel transitioning on some VCs, it's hardly there, instruments are still perfectly clear (I am talking about an almost not there effect compared to DX10 problems). However 32xS [Combined: 2x2 SS + 8x MS] + 8xSGSS blurs it with odd looking highlights, I feel I need glasses to view it, 4x SG and 2x just don't do anything, and worse still 8xS just does nothing to this micro-shimmer with 2xSG added. I couldn't live with 32xS + 8xSG detail missing and the performance suffered quite badly. SG applied when needed in a DX10 VC does not cause me to want to lean forward to view the instruments or cause a massive drop in performance, I had to zoom up quite close to read any numbers defeating the object. DX9 trees and fences were unaffected by even 32xS + 8xSG. I am concluding that it's not performing correctly, the loss of detail looks wrong, and makes me want to squint. I think this DX9 setup is problematic, there seems to be nothing to gain. DX10 does none of that, but DX10 performed far better than DX9 in all cases.

 

...forgot to mention seems odd to stack up three AA techniques whereas there are plenty of combined methods in the drop-down. Can't help feeling it would be in the dropdown if it was realistic.

You got a typo there, 8xS is 1x2SS + 4xMS according to Inspector. 

Anyway, Steve, there must be something wrong with your driver or Inspector installs if you see no effect of SGSS on AG shimmer. Have you tried deleting your FSX Inspector profile and creating a new one from scratch? maybe even a different driver?

 

 

...forgot to mention seems odd to stack up three AA techniques whereas there are plenty of combined methods in the drop-down. Can't help feeling it would be in the dropdown if it was realistic.

 

I guess SGSS is still in the transparency section because that's where it was supposed to be originally. It still runs as a transparency method, only that the alpha test always returns true. Besides, adding a 2xSGSS combination for each mode that supports 2x or more MS's, again for 4xSGSS & 8xSGSS...that's a lot of extra entries for the drop down.

 

I got some questions myself, like I don't get what the transparency MS methods are supposed to do... MS is a geometry mode that should ignore textures or shaders... yet there's a MS method for transparent textures? I'm missing something there.

Also if 8xS is just 4xMSAA, how can 8xSGSS run just fine coupled with 8xS? According to this 8xS is 8xMSAA though so who knows

Yes, fresh profiles many times.

 
But how about looking at it another way? Load up some aircraft in DX10 with simply 8xSSAA from the transparency section, give it a go. With the stock Baron first should see good if not exceptional results. Try some Carenados, or a BN2 and see the really bad AA in effect, it's like they are not AA'd at all. But then change the 8xSSAA to 8xSGSSAA, to see a marked difference - this is an obviously big difference. None the less there will be microshimmer when inspected very closely, (zoomed out and backed out to the limit of resolving the numbers on the dial).
 
Go back to DX9, with just 32xS see the same level of microshimmer, but I then add 8xSG and the detail is lost but to no real advantage to the view, the shimmer is less punchy, still can be seen, but the numbers are now too difficult to read. Like in the smaller picture of the corner fixing of the instrument I posted with/without. We are talking minute differences. Even so I have to zoom in surprisingly far to regain the detail. I've also been trying other screen resolutions, but the same results. Not closing the book on this just yet.

Yes, fresh profiles many times.

 

But how about looking at it another way? Load up some aircraft in DX10 with simply 8xSSAA from the transparency section, give it a go. With the stock Baron first should see good if not exceptional results. Try some Carenados, or a BN2 and see the really bad AA in effect, it's like they are not AA'd at all. But then change the 8xSSAA to 8xSGSSAA, to see a marked difference - this is an obviously big difference. None the less there will be microshimmer when inspected very closely, (zoomed out and backed out to the limit of resolving the numbers on the dial).

 

Go back to DX9, with just 32xS see the same level of microshimmer, but I then add 8xSG and the detail is lost but to no real advantage to the view, the shimmer is less punchy, still can be seen, but the numbers are now too difficult to read. Like in the smaller picture of the corner fixing of the instrument I posted with/without. We are talking minute differences. Even so I have to zoom in surprisingly far to regain the detail. I've also been trying other screen resolutions, but the same results. Not closing the book on this just yet.

 

Remember I can't test it, I'm an AMD user now (not sure why Avsim removed the PC config thingy in our user details). Not that I need to, I tested it on my former GTX's countless times.

The fact that your numbers are blurry and you can see the perf hit is a good indication that SGSS is working. I think at least we will agree to put to rest that theory that SGSS does nothing in DX9.

Yes, it does something, it's an effect, at 8x, but in DX10 there's no doubt it's SG.

 

edit: to be clear 4x and 2x did not produce a change, and I needed way more than 8xS.

...haha, sweet-talk me into agreeing with you? No. The SG setting is, after all, in the transparency section because it is for DX10 and above, and the effect in certain DX9 situations is just that, an effect.

...haha, sweet-talk me into agreeing with you? No. The SG setting is, after all, in the transparency section because it is for DX10 and above, and the effect in certain DX9 situations is just that, an effect.

 

What in the world makes you think TrAA doesn't apply to DX9? Especially when you just saw it in "effect", even if it might not be working properly on your end

 

http://aras-p.info/texts/D3D9GPUHacks.html#transpaa

haha, pretty unconvincing.

haha, pretty unconvincing.

 

No, it's an honest question Steve. I've asked you before and you ignored me. I'm just curious. What the technical reason?

https://takinginitiative.wordpress.com/2010/04/09/directx-10-tutorial-6-transparency-and-alpha-blending/

try changing sparse grid in DX10 with a normal model, makes no difference, but certain other models show up looking poorly AA'd without, for reasons touched on in that tutorial. I think there's more around on that site and others.

https://takinginitiative.wordpress.com/2010/04/09/directx-10-tutorial-6-transparency-and-alpha-blending/

try changing sparse grid in DX10 with a normal model, makes no difference, but certain other models show up looking poorly AA'd without, for reasons touched on in that tutorial. I think there's more around on that site and others.

 

Thanks for the link. At first sight I don't see anything related to D3D9 though

First of all the document in the "link" is provided for my clients. It's clear about what it does and there's nothing indeterminate about the completely repeatable results. It serves a specific purpose to get the job done. When I come to mucking around with DX9 and effects of messing around with different settings I find that 8xS does not cure any kind of shimmering, with or without SG settings chosen, so I said that was all at sea, it is. However I found 32xS + 8SG to have an "effect", you may like it you may not, you won't see it unless you look hard for it. That's not in the document because it wastes everyone's time and is misleading. Next when we look at DX10 and look at the shot I provided, that's no joke. With full SSAA there's no apparent AA applied on that model, until we set SG, where the huge marked obvious difference shows up. A "situation" arises where the transparency of the edge pixels of facets is determined differently during pre-rendering. That's all I can say on it, it's a DX10 and above thing.